Merge branch 'enhancement/improve_nodata_value_handling' into 'master'

Enhancement/improve nodata value handling

See merge request !3

geoarray/baseclasses.py

@@ -52,14 +52,17 @@ __author__ = 'Daniel Scheffler'
 
 
 class GeoArray(object):
+    """
+    This class creates a fast Python interface for geodata - either on disk or in memory. It can be instanced
+    with a file path or with a numpy array and the corresponding geoinformation. Instances can always be indexed
+    like normal numpy arrays, no matter if GeoArray has been instanced from file or from an in-memory array.
+    GeoArray provides a wide range of geo-related attributes belonging to the dataset as well as some functions for
+    quickly visualizing the data as a map, a simple image or an interactive image.
+    """
     def __init__(self, path_or_array, geotransform=None, projection=None, bandnames=None, nodata=None, progress=True,
                  q=False):
         # type: (Union[str, np.ndarray], tuple, str, list, float, bool, bool) -> None
-        """This class creates a fast Python interface for geodata - either on disk or in memory. It can be instanced with
-        a file path or with a numpy array and the corresponding geoinformation. Instances can always be indexed like
-        normal numpy arrays, no matter if GeoArray has been instanced from file or from an in-memory array. GeoArray
-        provides a wide range of geo-related attributes belonging to the dataset as well as some functions for quickly
-        visualizing the data as a map, a simple image or an interactive image.
+        """Get an instance of GeoArray.
 
         :param path_or_array:   a numpy.ndarray or a valid file path
         :param geotransform:    GDAL geotransform of the given array or file on disk
@@ -354,7 +357,7 @@ class GeoArray(object):
             if not self.is_inmem:
                 self.set_gdalDataset_meta()
             if self._nodata is None:
-                self._nodata = self.find_noDataVal()
+                self.find_noDataVal()
                 if self._nodata == 'ambiguous':
                     warnings.warn('Nodata value could not be clearly identified. It has been set to None.')
                     self._nodata = None
@@ -369,6 +372,9 @@ class GeoArray(object):
         # type: (Union[int, None]) -> None
         self._nodata = value
 
+        if self._metadata and value is not None:
+            self.metadata.global_meta.update({'data ignore value': str(value)})
+
     @property
     def mask_nodata(self):
         """
@@ -515,7 +521,7 @@ class GeoArray(object):
         if self._metadata is not None:
             return self._metadata
         else:
-            default = GDAL_Metadata(nbands=self.bands)
+            default = GDAL_Metadata(nbands=self.bands, nodata_allbands=self._nodata)
 
             self._metadata = default
             if not self.is_inmem:
@@ -705,6 +711,7 @@ class GeoArray(object):
         else:
             nodata = None
 
+        self.nodata = nodata
         return nodata
 
     def set_gdalDataset_meta(self):
@@ -736,12 +743,12 @@ class GeoArray(object):
             if 'nodata' not in self._initParams or self._initParams['nodata'] is None:
                 band = ds.GetRasterBand(1)
                 # FIXME this does not support different nodata values within the same file
-                self._nodata = band.GetNoDataValue()
+                self.nodata = band.GetNoDataValue()
 
             # set metadata attribute
             if self.is_inmem or not self.filePath:
                 # metadata cannot be read from disk -> set it to the default
-                self._metadata = GDAL_Metadata(nbands=self.bands)
+                self._metadata = GDAL_Metadata(nbands=self.bands, nodata_allbands=self._nodata)
 
             else:
                 self._metadata = GDAL_Metadata(filePath=self.filePath)
@@ -977,27 +984,27 @@ class GeoArray(object):
                 gdal.Unlink(out_path + '.aux.xml')
 
             elif self.metadata.all_meta:
-                    # set global domain metadata
-                    if self.metadata.global_meta:
-                        ds_out.SetMetadata(dict((k, repr(v)) for k, v in self.metadata.global_meta.items()))
+                # set global domain metadata
+                if self.metadata.global_meta:
+                    ds_out.SetMetadata(dict((k, repr(v)) for k, v in self.metadata.global_meta.items()))
 
-                    if 'description' in envi_metadict:
-                        ds_out.SetDescription(envi_metadict['description'])
+                if 'description' in envi_metadict:
+                    ds_out.SetDescription(envi_metadict['description'])
 
-                    # set band domain metadata
-                    bandmeta_dict = self.metadata.to_DataFrame().astype(str).to_dict()
+                # set band domain metadata
+                bandmeta_dict = self.metadata.to_DataFrame().astype(str).to_dict()
 
-                    for bidx in range(self.bands):
-                        band = ds_out.GetRasterBand(bidx + 1)
-                        bandmeta = bandmeta_dict[bidx]
-                        # meta2write = dict((k, repr(v)) for k, v in self.metadata.band_meta.items() if v is not np.nan)
-                        band.SetMetadata(bandmeta)
+                for bidx in range(self.bands):
+                    band = ds_out.GetRasterBand(bidx + 1)
+                    bandmeta = bandmeta_dict[bidx]
+                    # meta2write = dict((k, repr(v)) for k, v in self.metadata.band_meta.items() if v is not np.nan)
+                    band.SetMetadata(bandmeta)
 
-                        if 'band_names' in envi_metadict:
-                            band.SetDescription(self.metadata.band_meta['band_names'][bidx].strip())
+                    if 'band_names' in envi_metadict:
+                        band.SetDescription(self.metadata.band_meta['band_names'][bidx].strip())
 
-                        band.FlushCache()
-                        del band
+                    band.FlushCache()
+                    del band
 
             ds_out.FlushCache()
             del ds_out
@@ -1507,7 +1514,7 @@ class GeoArray(object):
     def get_subset(self, xslice=None, yslice=None, zslice=None, return_GeoArray=True,
                    reset_bandnames=False):
         # type: (slice, slice, slice, bool, bool) -> GeoArray
-        """Returns a new instance of GeoArray representing a subset of the initial one wit respect to given array position.
+        """Return a new GeoArray instance representing a subset of the initial one wit respect to given array position.
 
         :param xslice:          a slice providing the X-position for the subset in the form slice(xstart, xend, xstep)
         :param yslice:          a slice providing the Y-position for the subset in the form slice(ystart, yend, ystep)

geoarray/metadata.py

@@ -4,6 +4,7 @@ import os
 from pprint import pformat
 from copy import deepcopy
 from typing import Union  # noqa F401  # flake8 issue
+from collections import OrderedDict
 
 from geopandas import GeoDataFrame, GeoSeries
 import numpy as np
@@ -27,18 +28,22 @@ autohandled_meta = [
 
 
 class GDAL_Metadata(object):
-    def __init__(self, filePath='', nbands=1):
+    def __init__(self, filePath='', nbands=1, nodata_allbands=None):
         # privates
-        self._global_meta = dict()
-        self._band_meta = dict()
+        self._global_meta = OrderedDict()
+        self._band_meta = OrderedDict()
 
         self.bands = nbands
         self.filePath = filePath
         self.fileFormat = ''
+        self.nodata_allbands = nodata_allbands
 
         if filePath:
             self.read_from_file(filePath)
 
+        if nodata_allbands is not None:
+            self.global_meta.update({'data ignore value': str(nodata_allbands)})
+
     @classmethod
     def from_file(cls, filePath):
         return GDAL_Metadata(filePath=filePath)
@@ -73,8 +78,8 @@ class GDAL_Metadata(object):
 
     @band_meta.setter
     def band_meta(self, meta_dict):
-        if not isinstance(meta_dict, dict):
-            raise TypeError("Expected type 'dict', received '%s'." % type(meta_dict))
+        if not isinstance(meta_dict, (dict, OrderedDict)):
+            raise TypeError("Expected type 'dict'/'OrderedDict', received '%s'." % type(meta_dict))
 
         for k, v in meta_dict.items():
             if not isinstance(v, list):
@@ -83,11 +88,11 @@ class GDAL_Metadata(object):
                 raise ValueError("The length of the given lists must be equal to the number of bands. "
                                  "Received a list with %d items for '%s'." % (len(v), k))
 
-        self._band_meta = meta_dict  # TODO convert strings to useful types
+        self._band_meta = OrderedDict(meta_dict)  # TODO convert strings to useful types
 
     @property
     def all_meta(self):
-        all_meta = self.global_meta.copy()
+        all_meta = OrderedDict(self.global_meta.copy())
         all_meta.update(self.band_meta)
         return all_meta
 
@@ -190,8 +195,8 @@ class GDAL_Metadata(object):
         return 'Metadata: \n\n' + pformat(self.all_meta)
 
     def to_ENVI_metadict(self):
-        return dict(zip(self.all_meta.keys(),
-                        [self._convert_param_to_ENVI_str(i) for i in self.all_meta.values()]))
+        return OrderedDict(zip(self.all_meta.keys(),
+                               [self._convert_param_to_ENVI_str(i) for i in self.all_meta.values()]))
 
     def get_subset(self, bands2extract=None, keys2extract=None):
         # type: (Union[slice, list, np.ndarray], Union[str, list]) -> 'GDAL_Metadata'

geoarray/version.py

-__version__ = '0.8.13'
-__versionalias__ = '20190329.02'
+__version__ = '0.8.14'
+__versionalias__ = '20190329.03'